Biochemical behavior of N-oxidized cytosine and adenine bases in DNA polymerase-mediated primer extension reactions.

To clarify the biochemical behavior of 2'-deoxyribonucleoside 5'-triphosphates and oligodeoxyribonucleotides (ODNs) containing cytosine N-oxide (C(o)) and adenine N-oxide (A(o)), we examined their base recognition ability in DNA duplex formation using melting temperature (T(m)) experiments and their substrate specificity in DNA polymerase-mediated replication. As the result, it was found that the T(m) values of modified DNA-DNA duplexes incorporating 2'-deoxyribonucleoside N-oxide derivatives significantly decreased compared with those of the unmodified duplexes. However, single insertion reactions by DNA polymerases of Klenow fragment (KF) (exo(-)) and Vent (exo(-)) suggested that C(o) and A(o) selectively recognized G and T, respectively. Meanwhile, the kinetic study showed that the incorporation efficiencies of the modified bases were lower than those of natural bases. Ab initio calculations suggest that these modified bases can form the stable base pairs with the original complementary bases. These results indicate that the modified bases usually recognize the original bases as partners for base pairing, except for misrecognition of dATP by the action of KF (exo(-)) toward A(o) on the template, and the primers could be extended on the template DNA. When they misrecognized wrong bases, the chain could not be elongated so that the modified base served as the chain terminator.

INTRODUCTION

DNA bases are oxidized by n class="Chemical">reactive oxygen species (ROS) in the living body. Reactions of ROS with pyrimidine and purine bases result in the formation of several types of oxidation products in genomic DNA (1–7). For example, 5,6-dihydroxy-5,6-dihydrothymine (thymine glycol) and 5,6-dihydroxy-5,6-dihydrouracil (uracil glycol) are generated by the reactions of the hydroxyl radical, which is most mediated in the oxidation reactions, with pyrimidine bases. Additionally, the oxidation of guanine by singlet oxygen produces 8-oxo-7,8-dihydroguanine as a major oxidation product of guanine. These modifications have been found to block the chain elongation by DNA polymerases or to lead mutations (8–10). Oxidation by ROS occurs not only in genomic DNA, but also in the nucleotide pool. In fact, the reactions of dNTPs with ROS in the nucleotide pool were reported to produce oxidized dNTP derivatives, such as 8-oxo-7,8-dihydroguanosine 5′-triphosphate (11,12). The oxidized dNTP derivatives were incorporated into the growing chains on DNA templates using DNA polymerases in DNA replication and cause mutation or inhibition of chain elongation (13–15). Consequently, DNA base oxidation has been recognized as a major cause of genetic mutations that play an important role in many biological processes, such as cancers, aging, and genetic diseases (16–18).

Hydrogen peroxide, which is a more unreactive n class="Chemical">ROS than hydroxyl radical, is important in understanding the mechanism of oxidatively DNA damage. It is a major non-radical species produced by a series of metabolic processes of non-reactive superoxide anions, and is widely present in living cells (19–21). The main products obtained by the oxidation of cytosine and adenine bases with hydrogen peroxide are cytosine N-oxide (Co) and adenine N-oxide (Ao) derivatives, respectively (Figure 1) (22–32).

Figure 1.

Structures of Ao and Co derivatives.

Cadet reported that adenine N1-oxide is formed using n class="Chemical">hydrogen peroxide as a predominant ionic DNA base damage under non-radical conditions (29). 2′-Deoxyadenosine N1-oxide (3.2/106 bases) was detected to be present in the DNA of the untreated wild-type Proteus mirabilis cells (28). When they were exposed to nonlethal level of 10 mM hydrogen peroxide, the amount of this modified nucleoside was found to be 8.3 times higher than in the untreated cells (28). Upon exposure of calf thymus DNA to 10 mM hydrogen peroxide for 30 min, the number of the oxidized adenine residues was reported to be 0.7/104 bases. If this oxidative reaction of dCTP and dATP with hydrogen peroxide occurs in the nucleotide pool, the corresponding 2′-deoxyribonucleoside N-oxide 5′-triphosphates would be incorporated into DNA duplexes using DNA polymerases in DNA replication. Because these N-oxide derivatives have an additional oxygen atom at the Watson–Crick base pairing site, their hydrogen bond formation could be different from the normal ones. Therefore, there is a possibility that these 2′-deoxyribonucleoside N-oxides are mutagenic in DNA replication, such as the well-known 8-oxo-7,8-dihydroguanine (33). In addition, 2′-deoxyribonucleoside N-oxides might have a harmful effect on DNA transcription and replication.

It is known that 2′-deoxyribonucleoside n class="Chemical">N-oxide derivatives have some biological activities. Brown et al. (34) previously reported strong growth-inhibitory activity of adenosine N1-oxide and AMP N1-oxide in mouse sarcoma 180 cells. Adenine N1-oxide was converted into 2-hydroxyadenine by ultraviolet radiation (35). Its photochemical conversion might occur in skin as a rather minor process. In addition, it has been reported that adenosine N1-oxide was a highly selective inhibitor of vaccinia virus replication and viral late protein synthesis (36,37). Because 2′-deoxycytidine is readily oxidized to the N3-oxide derivative by treatment with hydrogen peroxide, it seems possible that the cytosine N3-oxide base forms in DNA upon treatment with hydrogen peroxide; however, this modified species has not been isolated from the oxidatively damaged DNA. The objective of this study is to examine the base recognition abilities of 2′-deoxyribonucleoside N-oxide moieties in DNA polymerase reactions to clarify their longstanding unknown biological properties.

In our previous study, we reported the synthesis of oligodeoxyribonucleotides (n class="Chemical">ODNs) containing 2′-deoxyribonucleoside N-oxide derivatives using a post-synthetic modification method (38). In this article, we report the base recognition abilities of Co and Ao bases by Tm experiments using synthetic ODNs incorporating them. We also report the substrate specificity of 2′-deoxyribonucleoside N-oxide derivatives in primer extension reactions using various DNA polymerases, i.e. not only in incorporation of dCoTP or dAoTP into DNA primers on DNA templates but also in incorporation of dNTP into DNA primers in DNA templates containing Co or Ao. In addition, the base recognition abilities of Co and Ao are also discussed on the basis of the hydrogen bond energies resulting from their ab initio calculations and MD simulations.

MATERIALS AND METHODS

General remarks

1H and n class="Chemical">31P NMR spectra were recorded at 270 and 109 MHz, respectively. The chemical shifts were measured from tetramethylsilane for 1H NMR spectra and 85% phosphoric acid (0 ppm) for 31P NMR spectra. UV spectra were recorded on a U-2000 spectrometer. High-performance liquid chromatography (HPLC) was performed using the following systems. Reversed-phase HPLC was done on a system with a 3D UV detector and a C18 column (4.6 × 150 mm). A linear gradient (0–30%) of solvent I [0.03 M ammonium acetate buffer (pH 7.0)] in solvent II (CH3CN) was used at 30°C at a rate of 1.0 ml/min for 30 min. DEAE–HPLC was performed on a Waters Alliance system with a Waters 3D UV detector and a Shodex DEAE column (8 × 75 mm). A linear gradient (0–45%) of solvent III [50 mM triethylammonium bicarbonate buffer (pH 7.2)] in solvent IV [500 mM triethylammonium bicarbonate buffer (pH 7.2)] was used at 25°C at a rate of 1.0 ml/min for 35 min. ESI mass was performed using MarinerTM (PerSeptive Biosystems Inc.). MALDI-TOF mass spectroscopy was performed using Bruker Daltonics [Matrix: 3-hydroxypicolinic acid (100 mg/ml) in H2O-diammoniumhydrogen citrate (100 mg/ml) in H2O (10:1, v/v)]. Highly cross-linked polystyrene (HCP) was purchased from ABI. DNA polymerase I Klenow fragment (exo−) was purchased from Promega. DNA polymerase I Klenow fragment (exo+), Pyrobest polymerase, and dNTPs were purchased from Takara Bio, Inc. Vent (exo−) DNA polymerase was purchased from New England Biolabs, Inc. ODNs used in the enzyme reactions and Tm experiments were purchased from Sigma-Aldrich Japan. 2′-Deoxyribonucleoside N-oxide 5′-triphosphates and ODNs containing the N-oxidized derivatives were synthesized, as described in Supplementary Data.

Tm experiments

ODNs n class="Chemical">containing 2′-deoxyribonucleoside N-oxides were synthesized by the post-synthetic oxidation method (38). An appropriate ODN (2 µM) and its complementary 2 µM ssDNA 14-mer or ssRNA 14-mer were dissolved in a buffer consisting of 1 M NaCl, 10 mM sodium phosphate and 0.1 mM EDTA adjusted to pH 7.0. The solution was maintained at 80°C for 10 min for complete dissociation of the duplex to single strands, cooled to 5°C at the rate of 0.5°C/min and heated to 80°C at the same rate using UV-1700™ (Shimadzu). During this annealing and melting, the absorption at 260 nm was recorded and used to draw UV melting curves. The Tm value was calculated as the temperature at which the first derivative of the UV melting curve had a maximum.

Single dNTP insertion reaction using Klenow fragment (exo−) DNA polymerase

The reaction mixture (10 µl) contained 50 mM Tris–HCl (pH 7.2), 10 mM n class="Chemical">MgSO4, 0.1 mM DTT, 100 nM 5′-FAM-labeled primer/template, 0.1 U enzyme and 10 µM dNTP (N = Ao, Co, A, G, C or T). The mixture was incubated at 37°C for 10 min, and the reactions were terminated by adding 30 µl of stop solution (98% formamide, 20 mM EDTA). After being vortexed gently, the samples were separated by electrophoresis using 20% denaturing polyacrylamide gel containing 7 M urea and were fluorescently visualized by Fujifilm FLA-2000G or FLA-7000. These results are shown in Supplementary Figures S1 and S3. The reactions using other enzymes were described in Supplementary Data.

Extension reactions of the primer with dNTPs in the presence or absence of dAoTP or dCoTP on templates composed of only natural bases or involving Ao or Co

Extension reactions were performed under the same conditions as described above for Klenow fragment (exo−) Dn class="Chemical">NA polymerase or Vent (exo−) DNA polymerase, except that the reactions were performed in the presence of three or four of dGTP, dTTP, dATP and dCTP (if necessary dAoTP, or dCoTP) (each dNTP at 10 µM). These results are shown in Figures 2 and 3 in the text. The results obtained when the reaction time was elongated to 30 or 60 min are also shown in Figures 2 and 3.

Figure 2.

Extension reactions with mixed dNTPs. (A) Sequences of 5′-FAM labeled 18-nt primer and 26-nt templates. (B) PAGE analysis of extension reactions using the natural and oxidized dNTPs (each dNTP at 10 µM) in the natural template. Lane 1 refers to the appropriate size markers. (C) Extension reactions in the presence of dGTP, dTTP, dATP and dCoTP (each dNTP at 10 µM) were shown in lanes 2 and 3 by KF (exo−) and lanes 6 and 7 by Vent (exo−). Extension reactions in the presence of dGTP, dTTP, dCTP and dAoTP (each dNTP at 10 µM) were shown in lanes 4 and 5 by KF (exo−) and lanes 8 anmd 9 by Vent (exo−). Lane 1 refer to the appropriate marker.

Figure 3.

Extension reactions with mixed dNTPs. (A) Sequences of 5′-FAM labeled 18-nt primer and 26-nt templates. (B) PAGE analysis of extension reactions using mixed dNTPs (each dNTP at 10 µM) in the oxidized templates. Lanes 1 refers to the appropriate marker. (C) PAGE analysis of extension reactions in the presence of dGTP, dTTP, dATP and dCTP (each dNTP at 10 µM). Lane 1 refer to the appropriate marker. Extension reactions by KF (exo−) are shown in lanes 2–5, and its reactions by Vent (exo−) are shown in lanes 6–9.

Extension reactions with mixed dNTPs. (A) Sequences of 5′-FAM labeled 18-nt primer and 26-nt templates. (B) PAGE analysis of extension reactions using the natural and oxidized n class="Chemical">dNTPs (each dNTP at 10 µM) in the natural template. Lane 1 refers to the appropriate size markers. (C) Extension reactions in the presence of dGTP, dTTP, dATP and dCoTP (each dNTP at 10 µM) were shown in lanes 2 and 3 by KF (exo−) and lanes 6 and 7 by Vent (exo−). Extension reactions in the presence of dGTP, dTTP, dCTP and dAoTP (each dNTP at 10 µM) were shown in lanes 4 and 5 by KF (exo−) and lanes 8 anmd 9 by Vent (exo−). Lane 1 refer to the appropriate marker.

Extension reactions with mixed dNTPs. (A) Sequences of 5′-FAM labeled 18-nt primer and 26-nt templates. (B) PAGE analysis of extension reactions using mixed n class="Chemical">dNTPs (each dNTP at 10 µM) in the oxidized templates. Lanes 1 refers to the appropriate marker. (C) PAGE analysis of extension reactions in the presence of dGTP, dTTP, dATP and dCTP (each dNTP at 10 µM). Lane 1 refer to the appropriate marker. Extension reactions by KF (exo−) are shown in lanes 2–5, and its reactions by Vent (exo−) are shown in lanes 6–9.

Steady-state kinetics methods in reactions using oxidized dNTPs and natural templates

A reaction mixture (10 µl) of 10 mM MgSO4, 0.1 mM n class="Chemical">DTT, 100 nM 5′-FAM-labeled primer/template duplex having the sequence [5′-FAM-(CGCGCGAAGACCGGTTAC)-3′/5′-(TAAGACXGTAACCGGTCTTCGCGCG)-3′, X = A, G, C and T], DNA polymerase Klenow fragment (exo−), dNTP (N = Ao, Co, A or C) in 50 mM Tris–HCl (pH 7.2) was incubated at 37°C for the appropriate times. The concentration of dNTP (0.2–50 µM) and the reaction time were adjusted for each reaction so as to give the product yield of ≤25%. The reactions were terminated by adding 30 µl of a stop solution (98% formamide, 20 mM EDTA). After being vortexed gently, the reactions were analyzed by electrophoresis with 20% denaturing polyacrylamide gel containing 7 M urea. The reactions were monitored with Fujifilm FLA-2000G. Relative velocity v was measured as the ratio of the extended product (Iext) to the remaining primer (Iprim) using v = Iext/Iprim t, where t represents the reaction time, and normalized for the lowest enzyme concentration used. The kinetic parameters (Km and Vmax) were obtained from Hanes–Woolf plots. Each parameter was averaged from three data sets. These results are shown in Table 1. Steady-state kinetics methods in reactions using natural dNTPs and oxidized templates were described in Supplementary Data.

Table 1.

Steady-state kinetic parameters for the insertion of single nucleotides into a template–primer duplex by Klenow fragment (exo−)

dNTP	Template	Km	Vmaxb	Efficiency
	(X)	(μM)	(percentage per minute)	(Vmax/Km)
Co	G	1.2 (0.30)	0.083 (0.016)	6.9 × 104
Co	A	53 (14)	0.006 (0.001)	1.0 × 102
Co	C	NDc	NDc	<1.0 × 102
Co	T	NDc	NDc	<1.0 × 102
C	G	0.25 (0.16)	0.51 (0.11)	2.0 × 106
Ao	T	1.2 (0.41)	0.060 (0.001)	4.9 × 104
Ao	G	15 (3.2)	0.059 (0.015)	3.8 × 103
Ao	C	14 (2.9)	0.008 (0.001)	5.3 × 102
Ao	A	NDc	NDc	<1.0 × 102
A	T	0.11 (0.020)	0.36 (0.027)	3.2 × 106

aAssay was carried out at 37°C for 0.55 min using 100 nM templateprimer duplex, 0.051 U enzyme and 0.250 µM dNTP in a solution (10 µl) containing 50 mM TrisHCl (pH 7.2), 10 mM MgSO4 and 0.1 mM DTT. The values in parenthesis are standard deviations.

bNormalized for the lowest enzyme concentration used.

cNot determined. No inserted products were detected after incubation for 10 min with 50 µM dNTP.

aAssay was carried out at 37°C for 0.55 min using 100 nM templateprimer duplex, 0.051 U enzyme and 0.250 µM dNTP in a solution (10 µl) n class="Chemical">containing 50 mM TrisHCl (pH 7.2), 10 mM MgSO4 and 0.1 mM DTT. The values in parenthesis are standard deviations.

RESULTS AND DISCUSSION

Tm experiment

To clarify the base pairing properties of the Co and n class="Chemical">Ao bases in DNA duplexes, Tm experiments were performed using duplexes formed between modified ODNs and their complementary ODNs with matched or single-mismatched sequences at the central position (Supplementary Table S1). An unmodified duplex of dA7GA6/dT6CT7 with a matched natural C–G base pair showed a Tm value of 47.1°C. In contrast, the Tm value of a duplex with an oxidized Co–G base pair decreased to 30.5°C with a significant drop of 16.6°C compared with the natural C–G matched base pair. In addition, the Tm values of DNA duplexes containing the mismatched pairs such as Co–T and Co–C also decreased. This result indicated that the base pairs of Co–T and Co–C were less stable by (ΔTm =) −3.3°C and −3.1°C than the canonical C–T and C–C mismatched base pairs, respectively. However, the DNA duplex containing a Co–A mismatched base pair showed a slightly decreased Tm value (ΔTm = −0.6°C) compared with that containing a C–A mismatch, which to our surprise, had the highest Tm value among the four base pairs. The DNA duplexes containing Ao exhibited a tendency similar to those having Co, as shown in Supplementary Table S1. The Tm value of a DNA duplex with an oxidized Ao–T base pair decreased by 11.5°C compared to that containing the natural A–T base pair, whereas the base pairs of Ao with G and C resulted in a decrease of 4.0–5.3°C in the Tm values of their duplexes. In the case of Ao, the combination of Ao with A gave in the highest Tm, similar to the Co–A base pair.

Next, similar Tm experiments were performed using duplexes formed between the modified n class="Chemical">ODNs and the complementary RNAs with a matched or single-mismatched sequence (Supplementary Table S2). The DNA–RNA duplexes having a Co–G or Ao–U base pair proved to be significantly destabilized, as described in the case of the modified DNA–DNA duplexes.

Synthesis of N-oxidized oligodeoxyribonucleotides and 2′-deoxyribonucleoside 5′-triphosphates

To examine the properties of 2′-deoxyribonucleoside N-oxide derivatives as substrates in the enzyme reactions, we synthesized dCoTP and dAoTP from dCTP and dATP, respectively. These modified 2′-deoxyribonucleoside 5′-triphosphates were easily obtained by treatment of dCTP or dATP with mCPBA in a mixed solvent of MeOH and NaHCO3 aqueous solution (Scheme 1). Consequently, these N-oxidation reactions proceeded very efficiently to yield the N-oxide derivatives as single products.

Scheme 1.

Synthesis of 2′-deoxycytidine and 2′-deoxyadenosine N-oxide 5′-triphosphates. Reagents and conditions: (a) mCPBA, aqueous NaHCO3-MeOH (1:1, v/v), rt, 6 h, 42% (for dCoTP), 48% (for dAoTP).

Synthesis of 2′-deoxycytidine and 2′-deoxyadenosine N-oxide 5′-triphosphates. Reagents and n class="Chemical">conditions: (a) mCPBA, aqueous NaHCO3-MeOH (1:1, v/v), rt, 6 h, 42% (for dCoTP), 48% (for dAoTP).

We also synthesized ODNs containing Co and Ao as templates in the enzyme reaction to clarify the base recognition abilities of the N-oxidized bases. These template ODNs containing Co and Ao were obtained by post-synthetic modification in 43% and 43% yields, respectively (Scheme 2) (38).

Scheme 2.

Synthesis of ODNs containing Co and Ao. Reagents and conditions: (a) mCPBA, MeOH, rt, 1 h, (b) concentrated NH3 aq, rt, 18 h, 43% (for ODN containing Co), 43% (for ODN containing Ao).

Single dNTP insertion by Klenow fragment (exo−)

To study the template-specific incorporation of these 2′-deoxyribon class="Chemical">nucleoside N-oxide 5′-triphosphates using DNA polymerase, we examined single nucleoside insertion reaction using the Klenow fragment (exo−) [KF (exo−)], which lacks the 3′- to 5′-exonuclease activity. The template incorporating one of the four natural bases (A, G, C and T) at position X was annealed with a 5′-FAM-labeled 18-nt primer in the presence of KF (exo−) and dCoTP or dCTP (Supplementary Figure S1A). After the enzymatic reaction, the products were analyzed by PAGE, as shown in Supplementary Figure S1B. Consequently, dCoTP was incorporated when X was G, as in the case of dCTP. Moreover, it was revealed that dCoTP was not incorporated at all when X was A, C or T. Similarly, the incorporation of dAoTP occurred at the opposite site of the parent T base in the template more precisely than that of dATP (lanes 6 and 8 versus 14 and 16). In the latter, wrong incorporation was observed when X was A or C. These results clearly suggested that ‘the Co and Ao bases could be incorporated accurately into the 3′-terminus of the primers according to the Watson–Crick base pairing rule even after N-oxidation’, although the above mentioned Tm experiments showed that these Co and Ao bases exhibited low base pairing abilities toward the original complementary G and T bases, respectively.

To investigate the incorporation efficiencies of the 2′-deoxyribon class="Chemical">nucleoside N-oxide 5′-triphosphates in more detail, we calculated steady-state kinetic parameters for the single insertion reaction using KF (exo−). The results are shown in Table 1. The Vmax/Km values in the primer-mediated incorporation of dCoTP at four opposite natural bases (A, G, C and T) in the templates were determined. As a result, the incorporation efficiency of dCoTP toward the opposite G in the template was found to be 690-fold higher than that of dCoTP toward the opposite A. This result shows that dCoTP is exclusively incorporated into the opposite site of G in the template. It was also suggested that if dCoTP was produced by the N-oxidation of dCTP in nucleotide pool, dCoTP having high fidelity would be rather selectively incorporated into elongating primers in a manner similar to that of dCTP in the replication. dAoTP also showed ∼13-fold higher selectivity for T than the other bases. The incorporation efficiencies of dCoTP or dCTP at the opposite site of G as well as dAoTP or dATP at the opposite site of T were extensively studied. Consequently, the Vmax/Km values of dCoTP, dCTP, dAoTP and dATP were calculated to be 6.9 × 104, 2.0 × 106, 4.9 × 104 and 3.2 × 106, respectively. Thus, the incorporation efficiency of oxidized dCoTP in the growing chain was ∼29-fold lower than that of dCTP. In addition, dAoTP was incorporated ∼65-fold less efficiently than dATP.

Single dNTP insertion reaction using other polymerases

To clarify the incorporation using other polymerases and the effect of their 3′- to 5′-exonuclease activity, we examined single insertion reactions using Klenow fragment (exo+) [KF (exo+)], Vent (exo−) polymerase and Pyrobest polymerase. Vent (exo−) polymerase and Pyrobest polymerase have been used as well known thermostable enzymes in PCR reactions (39). The former has no proofreading activity like KF (exo−), while the latter has a proofreading activity. The templates and primer used in this study and the results are shown in Supplementary Figure S2.

Supplementary Figure S2B shows the single dNTP-insertion reactions using Vent (exo−) polymerase. n class="Chemical">dCoTP and dAoTP were preferentially incorporated in a manner similar to that described in the reaction using KF (exo−). The PAGE image of the enzymatic reactions using KF (exo+) is shown in Supplementary Figure S2C. Although KF (exo+) has a 3′- to 5′-exonuclease activity, the 19-nt product containing dCo was explicitly observed when dCoTP was employed (lane 3). The control experiment of dCTP also showed a similar result. When dAoTP was used, it was incorporated at the opposite site of the T residue in the template to give the 19-nt product containing dAo (lane 13). As far as the substrate fidelity was concerned, we observed that dAoTP gave better results than dATP (lanes 10–12 versus 14–16). Regardless of the presence or absence of the exonuclease activity, this tendency was commonly observed in the case of KF (exo+) and KF (exo−), as mentioned above.

In sharp contrast to these results, the n class="Chemical">19-nt products containing the N-oxide bases were not observed in the single-insertion reactions using Pyrobest polymerase, as shown in Supplementary Figure S2D. In this case, it is likely that even when the N-oxide bases were incorporated into the 3′-terminus of the primer, they were removed by the excellent 3′- to 5′-exonuclease activity of Pyrobest polymerase (39). Otherwise, there is a possibility that this enzyme did not recognize the modified bases at all.

Primer extension reactions using mixed dNTPs involving dCoTP or dAoTP

We performed the full-length extension reactions using KF (exo−) and Vent (exo−) in the presence of a mixture of dNTPs involving n class="Chemical">dCoTP or dAoTP. The results are summarized in Figure 2B. The KF (exo−)-catalyzed primer extension using a mixture of dNTPs containing dCoTP on the template, where X was G, yielded mainly the one-base elongated 19-nt product that has dCo, but no full-length 25-nt product was observed. However, we observed an elongation product as a faint band at the position corresponding to 23-mer (lane 3). The formation of this product was explained in terms of consecutive addition of a CoGTAT sequence after the 3′-terminal site of the primer followed by termination at the second position (G) from the 5′-terminus of the template due to less effective incorporation of Co. Compared with these results, when X was T, the incorporation experiment with dAoTP gave a somewhat complicated PAGE pattern that was similar to the control without AoTP (lanes 6 and 7). This is because the part of the single strand on the template involves three T bases so that chain termination occurred at these positions (18, 21 and 24 nt) because of the poor incorporation efficiency of dAoTP (lane 6) and the absence of dAoTP (lane 7) after incorporation of the first Ao or misincorporation of the other bases. The elongated products in these primer extensions were more clearly observed when the reaction time was elongated from 10 to 60 min, as shown in Figure 2C. These results reflected that the incorporation efficiency of dAoTP was very poor.

In the case of Vent (exo−), the use of a mixture of dNTPs containing dCoTP (X = G) resulted in the formation of the one-nucleotide extended product as the minor band in PAGE (lane 9), while the use of a mixture of dNTPs containing dAoTP gave a result similar to that obtained by KF (exo−) (lanes 6 versus 12).

Therefore, it was concluded that both enzymes of KF (exo−) and Vent (exo−) allowed inn class="Chemical">corporation of dCoTP and dAoTP at the 3′-terminus of the primer. However, the presence of the once incorporated dCo or dAo residue, particularly the former, at the 3′-terminal site of the primer retarded significantly the incorporation of the next dNTP by the DNA polymerase. These properties could have a significant influence on the DNA replication process in living cells when dCoTP and dAoTP are produced by the oxidation of dCTP and dATP with H2O2 in the nucleotide pool.

Single-nucleotide-insertion reactions of dNTPs toward the opposite Co or Ao incorporated into DNA templates

Next, we performed the single-insertion reaction of n class="Chemical">dNTP toward the opposite Co or Ao in templates using DNA polymerases (Supplementary Figure S3). When X was Co, dGTP was selectively incorporated into the opposite position of Co in the template by use of KF (exo−) and Vent (exo−). Similar results were obtained when C was used as X. In addition, in order to examine the incorporation efficiencies of the natural dNTPs at the opposite N-oxidized base in the templates, we calculated the steady-state kinetic parameters of these reactions. The results are summarized in Table 2.

Table 2.

Steady-state kinetic parameters for the insertion of single nucleotide into a template–primer duplex by KF (exo−), and Vent (exo−)

Template base (X)	dNTP (N)	KF (exo−)a			Vent (exo−)b
		Km (μM)	Vmax (percentage per minute)	Efficiency (Vmax/Km)	Km (μM)	Vmax (percentage per minute)	Efficiency (Vmax/Km)
Co	G	0.68 (0.20)	0.18 (0.020)	2.7 × 105	2.0 (0.19)	0.81 (0.043)	4.1 × 105
Co	A	30 (2.5)	0.024 (0.002)	8.2 × 102	8.3 (1.1)	0.017 (0.001)	2.0 × 103
Co	C	NDc	NDc	<1.8 × 102	NDc	NDc	<1.5 × 103
Co	T	NDc	NDc	<1.8 × 102	NDc	NDc	<1.5 × 103
C	G	0.031 (0.012)	0.92 (0.056)	3.0 × 107	0.084 (0.018)	0.55 (0.029)	6.6 × 106
Ao	A	6.6 (0.94)	0.061 (0.002)	9.2 × 103	12 (2.9)	0.019 (0.001)	1.5 × 103
Ao	T	61 (13)	0.019 (0.001)	3.2 × 102	2.0 (1.4)	0.059 (0.007)	2.9 × 104
Ao	C	130 (28)	0.024 (0.001)	1.8 × 102	NDc	NDc	<1.5 × 103
Ao	G	NDc	NDc	<1.8 × 102	NDc	NDc	<1.5 × 103
A	T	−	−	−	0.45 (0.17)	2.05 (0.49)	4.6 × 106

aAssay was carried out at 37°C for 0.25–10 min using 100 nM template–primer duplex, 0.05 U enzyme and 0.1–500 µM dNTP in a solution (10 µl) containing 50 mM Tris–HCl (pH 7.2), 10 mM MgSO4 and 0.1 mM DTT.

bAssay was carried out at 74°C for 0.25–10 min using 100 nM template–primer duplex, 0.1 U enzyme and 0.05–50 µM dNTP in a solution (10 µl) 20 mM Tris–HCl (pH 8.8), 10 mM (NH4)2SO4, 10 mM KCl, 2 mM MgSO4 and 0.1% Triton X-100. The values in parenthesis are standard deviations.

cNot determined. No inserted products were detected after incubation for 10 min with 1000 µM dNTP.

The incorporation efficiency of n class="Chemical">dGTP at the opposite Co in the reaction using KF (exo−) indicated at least ∼330-fold higher value than those of the other dNTPs. Moreover, in the reaction using Vent (exo−), its incorporation efficiency was at least ∼205-fold higher than those of the other dNTPs at the opposite Co. Therefore, it was found that dGTP was rather selectively chosen from the four natural dNTPs as the complementary base for Co. The Vmax/Km values of dGTP at the opposite C and Co were 3.0 × 107 and 2.7 × 105, respectively, in the case of KF (exo−). The incorporation efficiency of dGTP at the oxidized Co was significantly lower by two orders than that of dGTP at the natural C.

On the other hand, we obtained interesting results in the insertion reaction using the template having Ao. When Vent (exo−) was employed, n class="Chemical">dTTP was predominantly incorporated into the site opposite Ao over dTTP in a manner similar to that of the natural A. To our surprise, when KF (exo−) was used, dATP was more preferentially incorporated into the opposite position of Ao in the template than dTTP. The Vmax/Km value of dATP (9.2 × 103) was ∼30-fold higher than that of dTTP (3.2 × 102), as shown in Table 2. Therefore, if Ao is produced by the N-oxidation of the adenine base in DNA genome or Ao is incorporated into DNA genome by AoTP-mediated replications, an Ao–A mismatch base pair might be formed in the DNA replication process. Its mismatch base pair would cause a significant transversion mutation to induce genetic diseases.

In addition, we carried out the insertion reactions using KF (exo+) and Pyrobest (Supplementary Figure S3). KF (exo+) incorporated not only n class="Chemical">dGTP as a match substrate, but also dATP as a mismatch substrate into the site opposite Co. In the template containing Ao, similar mis-incorporation of dATP was observed. Considering the results using KF (exo−), Klenow fragment tends to have an ability of incorporation of dATP into the opposite site of N-oxide bases in the template. In the reactions using Pyrobest, dGTP and dTTP were slightly incorporated into the opposite Co and Ao, respectively.

Primer extension reactions with mixed dNTPs

Next, we performed the full-length extension reactions using KF (exo−) or Vent (exo−) in the presence of various mixtures of n class="Chemical">dNTPs and the templates containing Co or Ao, as shown in Figure 3B. In the case of KF (exo−), the extension of the primer was stopped after the incorporation of dGTP and dATP at the opposite Co and Ao, respectively, in the template (lanes 3 and 6). Thus, the 2′-deoxyribonucleoside N-oxide residues in the templates showed very strong inhibition in the elongation of the primer by KF (exo−). This inherent property would interrupt the DNA-replication process, and thus prevent precise genetic information from transferring to new genomic DNA.

In contrast, the enymatic reactions using Vent (exo−) showed interesting results different from those derived from KF (exo−). It turned out that Vent (exo−) n class="Chemical">could extend the primer over Co or Ao in the template to give elongated primers, although these reactions generated shorter extension products than the full-length products (lanes 9 and 12).

We also examined in more detail the extension reactions under the conditions of prolonged reaction time to clarify whether longer products were generated (Figure 3C). As the result, the prolonged reaction using the template n class="Chemical">containing Ao gave the 25-nt full length DNA oligomer as the major product. Using the template containing Co, we observed a 24-nt main product probably lacking the A residue at the 3′-terminal position of the primer. It is unclear why such a truncated oligomer was formed.

Therefore, we found that the enzymatic behavior of 2′-deoxyribonucleoside n class="Chemical">N-oxide derivatives in the template varied depending on the type of DNA polymerases. In the case of KF (exo−), the N-oxide bases in the template strongly inhibit the extension of the primer. In the case of Vent (exo−), the primer elongates beyond the N-oxide bases on the template without stopping at the next site from the oxidized damage site.

Ab initio calculations of putative Co–G and Ao–T base pairs

To discuss the base recognition ability of the modified bases of n class="Chemical">Co and Ao from a theoretical viewpoint, we calculated the hydrogen bonding ability at the Watson–Crick base pairing site of Co and Ao. The ab initio calculations of the oxidized and natural base pairs at the MP2/6-31+G**//HF/6-31+G** level indicated that the hydrogen bonding geometries of the oxidized Co–G and Ao–T base pairs were different from those of the natural C–G and A–T base pairs, respectively (40–43).

Interestingly, the optimized structure of the Co–G base pair suggested that the n class="Chemical">oxygen atom of the N-oxide moiety can form two hydrogen bonds with the NH1 and NH2 of the guanine base (Supplementary Figure S4A). It was also suggested that this unique Co–G base pair has a hydrogen bond energy of −23.6 kcal/mol that is only a little smaller by 2.2 kcal/mol than that (−25.8 kcal/mol) of the C–G base pair. The oxidized Ao–T base pair has two hydrogen bonds, and the hydrogen bond energy shows a minimal change of 0.4 kcal/mol (Supplementary Figure S4B).

We also found that the oxidized base pairs have slightly longer C1′–C1′ distances compared to those of the normal base pairs. The C1′–C1′ distances of the n class="Chemical">Co–G and Ao–T base pairs were 12.3Å and 12.6Å, respectively, whereas the C1′–C1′ distance of the canonical base pair was 10.7Å. As the increase in the C1′–C1′ distance was expected to disturb the regular B-type backbone structure, it would be a major cause of the significant decrease in the Tm value observed by N-oxidation at the center position in the ODNs, as shown in Supplementary Figure S1. In addition, in the MD simulations of DNA duplexes containing the oxidized base pair, we confirmed that the presence of the unusual base pair resulted in significant distortion of DNA duplexes (Supplementary Figures S5 and S6). The average base pair forms obtained by the MD calculation were very similar to those resulted from the ab initio calculation of the model base pairs. In the DNA polymerase reaction, the increased C1′–C1′ distance would be crucial for the low-incorporation efficiency of the N-oxide derivatives. Additionally, the inhibition of chain elongation after the incorporation of dCoTP might be due to the alteration of the base pairing geometry which would influence substrate recognition in the active site of DNA polymerases. However, in this enzymatic reaction, dCo and dAo could be selectively recognized as the complementary G and T bases at the opposite positions at the 3′-terminal site of the primer. This is probably due to the flexible 3′-terminal site that spatially allows the formation of the wider oxidized base pairs, as shown in Supplementary Figure S4.

CONCLUSION

This study revealed that when dC and dA in Dn class="Chemical">NA–DNA or DNA–RNA duplexes were replaced with dCo and dAo, respectively, their thermal stability decreased significantly. From a series of Tm experiments of modified DNA–DNA (or RNA) duplexes, it was also disclosed that these N-oxide bases did not show selective binding ability to any canonical nucleobase at the opposite site in the complementary strand in modified DNA–DNA duplexes. To our surprise, however, the primer extensions using dCoTP and dAoTP in the presence of KF (exo−) showed markedly selective incorporation of dCo and dAo into the opposite site of dG and dT, respectively, on the template DNA, as shown by the PAGE analysis. We measured single-nucleotide incorporation kinetics for KF (exo−)-catalyzed primer extension to clarify the substrate specificity in more detail. From the data of the steady-state kinetic parameters for the single-nucleotide insertions, dCoTP exhibited rather high-incorporation selectivity toward G in the template with at least 690-hold fidelity compared those toward the other bases. On the other hand, dAoTP showed 13-hold fidelity toward T compared with the other bases. It was found that the incorporation efficiencies of dCoTP and dAoTP decreased 29 and 65 times more than those of the unmodified dCTP and dATP, respectively.

The use of Vent (exo−) and KF (exo+) gave results similar to those obtained by KF (exo−) as far as the specific incorporation of the modified bases was n class="Chemical">concerned. Interestingly, Pyrobest DNA polymerase did not give single-nucleotide insertion products when dCoTP and dAoTP were used as the substrates. Probably, this result would be due to the strong 3′- to 5′-exonuclease activity of this enzyme.

The experiments using a mixture of dNTPs involving n class="Chemical">dCoTP and dAoTP suggested that chain extension could occur from the N-oxide bases once introduced into the 3′-terminus of the primer, but the elongation was significantly retarded at the modified position.

We also synthesized ODNs n class="Chemical">containing dCo or dAo according to our original method. These synthetic materials enabled us to determine if dNTPs can serve as the substrates in DNA-polymerase-mediated reactions when the modified bases were mis-incorporated into DNAs by duplication with dCoTP and dAoTP or by direct oxidation of DNAs. As a result, dGTP was found to be rather selectively chosen from the four natural dNTPs as the complementary base for Co in primer extensions using KF (exo−) and Vent (exo−) polymerases, but further chain elongation was greatly suppressed in the presence of four dNTPs. In particular, it is noteworthy that in the case of KF (exo−), dATP was more preferentially incorporated into the opposite position of Ao in the template than dTTP, but this misincorporation inhibited the next incorporation at the next position. A longer primer chain was observed using this enzyme upon prolongation of the reaction time when Co or Ao was located in the template DNA. More interestingly, extensive chain elongation was observed when the Ao base was located in the template DNA and Vent (exo−) polymerase was employed.

From the theoretical point of view, the Co–G and n class="Chemical">Ao–T base pairs could be formed as the most stable structures that have sufficient hydrogen bond energies. Molecular dynamics simulation also suggested DNA duplexes incorporating these modified bases could be kept as stable double strands although some part of the duplexes was disordered. Probably, since the oxidation of the parent bases allowed the smallest modification on the bases and the hydrogen bonding acceptor site shifted only by a distance of the N–O bond, similar Watson–Crick type hydrogen bonds can be formed between Co or Ao with the complementary bases in an enzyme–primer–template complex.

Precise base recognition abilities of the n class="Chemical">N-oxidized bases suggest that Co and to a lesser extent Ao are not highly mutagenic and only weakly lethal lesions. In contrast, 2′-deoxyribonucleoside N-oxide derivatives also have the property to inhibit the elongation of primer in DNA polymerase-mediated reactions. Therefore, it is believed that this property of 2′-deoxyribonucleoside N-oxide derivatives may induce serious errors in DNA replication if these oxidized derivatives are generated in living cells.

To clarify more biological behavior of 2′-deoxyribonucleoside n class="Chemical">N-oxide derivatives, it would be necessary, for example, to perform shuttle vector experiments that would involve the transfection of host cells with N-oxide base containing plasmids. The biochemical studies are now under way and will be reported elsewhere.

SUPPLEMENTARY DATA

Supplementary Data are available at NAR Online.

FUNDING

Grant-in-Aids for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology, Japan; Health Sciences Research Grants for Research on Psychiatric and Neurological Diseases and Mental Health from the Ministry of Health, Labor and Welfare of Japan, and by the global n class="Chemical">COE project (in part). Funding for open access charge: Grant-in-Aids for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology, Japan.

Conflict of interest statement. None declared.